pyglpi

This is thin wrapper around the GLPI REST API using Hammock and aiding in session creation, iterating through result ranges and translating search options between readable and persistent UIDs and the IDs used by the API.

Registering your application with GLPI

Register the application you are using the API from in your GLPI instance under Setup → General → API to receive the app_token.

Invocation

The most basic invocation looks like this:

glpi = pyglpi.GLPI()

This assumes that the URL to your GLPI instance is stored in the environment variable GLPI_URL, and your app token in GLPI_APP_TOKEN.

You can also pass them via arguments:

glpi = pyglpi.GLPI(
    url='https://glpi.example.org/apirest.php',
    app_token='...',
)

For all following examples, we will assume environment variable usage where applicable.

Login

There are several ways to login, that are tried in this order:

1. Login via user credentials
2. Login via user token
   1. passed via code
   2. passed via environment variable
3. Unauthenticated usage

Login via user credentials

Use this only when the API is used with different, user-provided credentials.

glpi = pyglpi.GLPI(
    credentials=('username', 'password'),
)

Login using user token

You can generate the user_token in your profile. This is the recommended way to authenticate to GLPI if you have to store the credentials.

You can then login like this:

glpi = pyglpi.GLPI(
    user_token='0123456789abcdef0123456789abcdef01234567',
)

Instead of passing the user token via argument in code, you can also set it via the environment variable GLPI_USER_TOKEN and simply omit it in the invocation:

os.environ['GLPI_USER_TOKEN'] = '0123456789abcdef0123456789abcdef01234567'
glpi = pyglpi.GLPI()

No login

There are very few API endpoints that don't require login, so the code only makes unauthenticated calls if none of the above methods were used.

Calling API methods directly / Paged replies

All API endpoints are callable as usual with Hammock. However, for GET requests, the special ranges attribute is added to the response to aid in iterating through paged responses:

for result_range in glpi.Computer.GET().ranges:
    for item in result_range.json():
        print(item['name'])

Search

There is a special helper function to convert the search criteria from a python datastructure to the correct GET parameters, on-the-fly translating field names to the numeric search option IDs that the API expects, and translating the result keys back to the readable UIDs.

criteria = [{
    'field': 'name',
    'searchtype': 'contains',
    'value': '.example.net$',
}]

for item in pyglpi.search(glpi, 'Computer', criteria):
    print(item['name'])

Note that while in the input parameter criteria the search function accepts both variants Computer.name and just name, as well as the plain numeric ID 1, the result will always have the itemtype stripped from the full UID as returned by searchOptions API endpoint, to be at least similar to the fetching of items (see above).

Also note that ranges are automatically iterated through inside the helper function.

Search for AllAssets

The API supports searching for all kinds of assets in a single query. However, because AllAssets cannot be passed to the searchOptions endpoint, we have to use the IDs from some other itemtype. This is supported by passing in the search_options to use for translation:

criteria = [{
    'field': 'name',
    'searchtype': 'contains',
    'value': '.example.net$',
}]
search_options = glpi.searchOptions.Computer.GET().json()

for item in pyglpi.search(glpi, 'AllAssets', criteria, search_options):
    print(item['name'])